Here’s a concise forensic-style summary of the analysis:
? Forensic Analysis Summary
Image Source: Social media download (metadata stripped)
Objective: Determine if AI-generated or real photo
Analyses Performed
- Sensor Noise Analysis
- Mean ≈ 0 (no bias)
- Std dev ≈ 13.9 (normal sensor noise)
- Noise strongest in shadows/edges → consistent with real cameras.
- Frequency Spectrum
- Smooth natural falloff in high frequencies.
- No grid-like or spiked artifacts (common in AI).
- Local Texture Variance
- Wide spread of patch variances (mean ≈ 1184, std ≈ 1574).
- Mix of smooth (skin, wall) and detailed (clothing, bricks) areas.
- Distribution looks natural, unlike AI’s uniform/chaotic patterns.
- Fine-Detail Edge Map (Laplacian)
- Strong, clean edges at boundaries.
- Consistent textures with no warped or “melting” artifacts.
- High kurtosis (~26), indicating realistic edge vs. smooth area balance.
Conclusion
All forensic tests (noise, frequency, texture, edges) indicate the image is highly likely a real camera photograph.
No clear evidence of AI generation was found.
